Intermittent asymmetric connectivity between Telia Finland and Azure - possible routing/peering issue?

Question

Intermittent asymmetric connectivity between Telia Finland and Azure - possible routing/peering issue?

Antti T 0

I use several Azure services for both work and personal purposes, and lately connectivity from Telia Finland consumer connections to Azure has been very unreliable.

A group of Telia Finland customers in different parts of Finland have compared incidents and found that the same brief outages occur for all of us at roughly the same time, typically lasting 5–30 seconds.

During these incidents, I have taken simultaneous packet captures on both ends. Packets sent from the Telia connection toward the Azure VM are visible in the capture on the Azure side, while packets sent from Azure back toward the Telia connection are never seen in the capture on the Telia side. This shows that the failure is asymmetric and affects the Azure → Telia direction.

If I route the same traffic through a VPN using another provider, connectivity remains stable. This strongly suggests that the issue is somewhere on the return path between Microsoft/Azure and Telia Finland, potentially involving routing or peering rather than an issue with an individual customer connection.

I have contacted Telia, but they say the issue is not on their side and have not been willing to investigate further.

Unfortunately, I cannot troubleshoot the Azure-side path very effectively myself. Traceroute from my Azure VMs does not provide useful results, Azure does not appear to offer a public looking glass, and I do not have a support plan that allows me to open a networking ticket.

Is there a way to get this investigated by the Azure networking team, or to obtain routing/peering diagnostics from the Azure side?

If useful, I can provide timestamps of affected periods, source/destination IP addresses etc.

0 comments

1 answer

Your answer

Answer 1

Hi @Antti T

The simultaneous packet captures are particularly useful here. If packets from Telia reach the Azure VM during the incident, but the return packets from Azure aren't observed on the Telia side, that points to an asymmetric return-path problem somewhere beyond the VM, rather than a straightforward inbound NSG or guest OS issue.

The fact that the problem disappears when the same client traffic is routed through another VPN/provider strengthens that possibility. However, packet captures alone can't establish whether the loss occurs within Microsoft's network, at Microsoft/Telia peering, or within Telia's return path.

I would collect evidence from Azure Network Watcher during both a working period and one of the 5–30 second failure windows. In particular, use Connection troubleshoot, packet capture, and Next Hop/effective-route information. Network Watcher can test connectivity to an external IP and identify the Azure-side next hop and routing information.

For escalation, keep the following for several incidents:

Exact UTC timestamps
Azure VM public IP and affected Telia source public IPs
Source/destination ports and protocol
Simultaneous packet captures from both endpoints
Working versus failing traceroute/MTR results where available
Azure region and VM/resource IDs
Network Watcher Connection Troubleshoot results

Unfortunately, detailed Microsoft backbone/BGP/peering telemetry isn't exposed through Network Watcher or a public Azure looking glass. If the Azure diagnostics show no VM/VNet-side routing or filtering problem, this will likely require Microsoft and Telia to investigate their respective routing/peering telemetry using the timestamps and IP pairs.

One complication is support entitlement: Microsoft states that technical Azure support normally requires a support plan; free support covers subscription-management matters rather than normal technical networking investigations.

Given that multiple Telia customers in different locations are apparently experiencing the outage simultaneously, I'd also ask Telia to escalate this to their network operations/peering team, providing the packet captures and exact UTC timestamps rather than treating it as an individual broadband fault.

I wouldn't make any Azure routing changes yet. The evidence you've gathered is much more valuable for correlating the failure against Microsoft/Telia routing telemetry.

Sharing these references with you:

Azure Network Watcher – Troubleshoot outbound connections

Diagnose VM routing with Network Watcher

Create an Azure support request

Please "Accept the Answer" if this information helped you. This will help us and others in the community.

Antti T 0 Reputation points

2026-08-21T16:42:33.8533333+00:00

Sadly Telia Finland is not very cooperative in getting this resolved.

The tickets I opened to them seem to end in a resolution of "not Telia problem" even though testing proves Microsoft+Telia is the only common factor here.

So far I guess the recommendation I have to start giving is that if you want to host a server with good connectivity requirements such as a game server to people in Finland then avoiding Microsoft Azure is a good idea because Telia<=>Azure connectivity is just not reliable and Telia is a widely used ISP in Finland with consumers often not having a proper alternative...
Allan Solomon Mejia 5,420 Reputation points

2026-08-21T18:29:33.9033333+00:00

Hello @Antti T

I understand the frustration, especially when you've reproduced the issue across multiple Telia connections, and the tickets keep being closed.

I would hesitate, though, to conclude that Azure itself is generally unsuitable for connectivity-sensitive workloads based on this case. The evidence appears to show a specific network path/peering issue between Telia and Microsoft, rather than an availability problem with the Azure-hosted service itself.

At this point, since Telia isn't progressing the investigation, I would open/escalate the case with Microsoft Azure Networking Support as well and provide the traceroutes/MTRs from affected Telia connections, along with equivalent tests from working ISPs, exact UTC timestamps, the destination public IP, and the Telia ticket numbers.

Microsoft can then examine the Microsoft-side routing/peering telemetry and, if necessary, engage the appropriate peering team. Having cases open on both sides of the interconnection gives you a much better chance of correlating this than continuing to troubleshoot the Azure VM itself.

Your comparison showing Telia → Azure fails while other ISP → same Azure endpoint works is particularly useful evidence. I would preserve those results rather than change the Azure environment.

Hopefully Microsoft and Telia can use that data to determine exactly where the path is breaking.

Help make this community better for everyone: if this answer resolved your issue, please accept it or upvote it. If not, share more details in a comment so we can continue the discussion and find the right solution.